• Default precision changed to double
  • regress now supports multiway clustering of standard errors
  • Dofiles come with variable autocompletion (only variables currently in dataset, no parsing of gen x = ... yet it seems)
  • Missing no longer considered a very large number
  • Running probit models now comes with a warning that these are not regular coefficients
  • Seamless integration with R
  • Dofiles with less than 20% of lines devoted to comments require a confirmation step to save
  • New command: import smart filename uses ML technology to determine all import parameters, just need to specify the file
  • Reshape now 100x faster
  • Name changed to STATDA in the UK, STAYTA in the US (all caps)
  • Push for more Bayesian statistics: by default all regressions are now run with the bayes prefix, to get classic estimates you must use the outdated: prefix
Happy April 1st everyone!